You are on page 1of 17

Emotional and Behavioural Difficulties, 2014

Vol. 19, No. 2, 210–225, http://dx.doi.org/10.1080/13632752.2013.844414

Bullying by definition: an examination of definitional components of


bullying
Susan Goldsmida* and Pauline Howieb
a
Centre for Mental Health Research, The Australian National University, Canberra, Australia;
b
School of Psychology, The University of Sydney, Sydney, Australia

Lack of definitional consensus remains an important unresolved issue within bullying


research. This study examined the ability of definitional variables to predict overall level
of victimisation (distress, power inequity, and provocation as predictors) and bullying
(intention to harm, power inequity, and provocation as predictors) in 246 Australian
university students. All variables were measured using the Victimisation and Bullying
Inventory (VBI), with behaviour assessed separately for tertiary institution, workplace
and home contexts. Regression analysis revealed that, as expected, higher levels of
distress predicted higher levels of victimisation (in all contexts) and higher levels of
intention to harm predicted higher levels of engagement in bullying (in work and home
contexts). Challenging definitional theory, bullying was reported as most commonly
occurring between two equals, from both the victim and bully perspective, and individ-
uals who bullied others blamed the victim for provoking the behaviour twice as often
as victims felt that they had provoked it.
Keywords: bullying; victimisation; definition; power inequity; provocation; distress;
intention to harm

Introduction
Despite general agreement that bullying is a common and significant problem, one of the
most confronting research challenges is the achievement of definitional consensus (Parada
2006; Sanders 2004). Without definitional consensus, bullying research has lacked focus
and direction, with a range of aspects of the phenomenon being sampled, reflecting differ-
ences between researchers in their conceptualization of bullying (Griffin and Gross 2004).
The aim of the current study was to conduct an empirical examination of the definitional
elements of bullying and victimisation, in order to provide an objective empirical basis for
definition, using an early post-school sample.
A commonly cited and influential definition of bullying was provided by Olweus
(1993a): ‘ . . . a student is being bullied or victimized when he is exposed repeatedly and
over time to negative actions on the part of one or more other students’ (9). This defini-
tion has been expanded or modified by a number of researchers, to emphasise different
aspects of the bullying phenomenon, such as the presence of a power inequity (Salmivalli
and Nieminen 2002), or to highlight the different forms that bullying can take (Mynard
and Joseph 2000) as more has become known about its nature (Parada 2006). Although the
majority of definitions refer specifically to bullying in school contexts, similar definitions

*Corresponding author. Email: susan.goldsmid@anu.edu.au

© 2013 SEBDA
Emotional and Behavioural Difficulties 211

have been applied to adult bullying. A literature review of definitions across a wide range
of ages reveals that several criteria are common to most. These will now be described.

Repetition
The vast majority of definitions refer to the fact that the behaviour must be repeated in order
to be categorized as bullying (e.g. Baldry and Farrington 2004; Connolly and O’Moore
2003; Olweus 1993b; Roland 2002; Salmivalli and Nieminen 2002). This serves to dis-
tinguish bullying from single aggressive acts, and to emphasize that bullying not only
produces immediate distress and harm, but also the threat of future attack (Lee 2006).
However, there is no general consensus on the frequency of behaviour needed to satisfy
this criterion (Cowie et al. 2002). In response to this, researchers have begun advocating
that victimisation and bullying should be conceptualized as continua, as any frequency cut
off imposed is arbitrary (Marsh et al. 2004). As a result, research in schools (e.g. Marsh
et al. 2004) and workplaces (e.g. Einarsen, Hoel, and Notelaers 2009) has adopted the
practice of measuring victimisation and bullying in terms of a combination of both the
frequency and variety of behaviours experienced in a given time period.

Victimisation distress
Whilst not explicitly stated in most definitions of bullying, there is an overriding assump-
tion that experiencing victimisation causes distress in both children and adults, ranging
from mild annoyance to severe psychological, social and physical trauma (Gregor 2004).
The absence of distress in definitions of bullying may be due to the subjective nature of the
judgment of distress and the practical difficulties of applying this criterion in measurement
instruments, given that responses to comparable levels of victimisation may vary greatly
due to individual vulnerabilities (Rigby 1997). Despite this, the consistent finding of asso-
ciations between distress and victimisation suggests the importance of consideration of this
criterion.

Intention to harm
This criterion is theoretically important in defining bullying, as it distinguishes bully-
ing from accidental, and therefore unintended harm (Anderson and Bushman 2002).
The majority of definitions of bullying contain terms such as ‘deliberate’ (Baldry and
Farrington 2004; Salmivalli and Nieminen 2002). In practice, however, the criterion of
intention is problematic, as it assumes sufficient self-awareness and honesty on the part
of the bully. It also involves a highly subjective judgement, on which the bully, the vic-
tim, and observers may all differ. However, the importance of distinguishing bullying from
accidental harm means that this criterion must be retained.

Power inequity
This is a common definitional criterion, based on the argument that it is the illegitimate
use of power by one person over another that distinguishes bullying from violence or gen-
eral aggression (Baldry and Farrington 2004; Connell and Farrington 1996; Mynard and
Joseph 2000; Rigby 2001; Salmivalli and Nieminen 2002; Slee and Rigby 1993; Smith
212 S. Goldsmid and P. Howie

and Brain 2000). Research in both adults and children, has reported that this power imbal-
ance can be physical, psychological, economic or hierarchical (Connell and Farrington
1996; Olweus 1993a, 1993b; Rigby 2001; Smith and Brain 2000). However, a number of
researchers have opposed this criterion (e.g. Connell and Farrington 1996; Parada 2006),
as targets of bullying often fail to report inequity. For example, Ireland and Ireland (2003)
found that, in incarcerated male offenders, 52% viewed bullies as being only ‘sometimes’
more powerful than their victims and 22% reported that bullies were not more powerful.
As omitting this criterion could misrepresent bullying, and as it is unclear from research
to date whether participants are failing to recognize an inequity that objectively exists, it
is important to examine further the nature of power imbalance in bullying situations in a
range of populations.

Provocation
Bullying is generally proposed to form part of a sub-set of aggression known as proactive
aggression, that is, the use of unprovoked aversive means to influence or coerce another per-
son (Griffin and Gross 2004; Salmivalli and Nieminen 2002). However, some researchers
argue that the justification of behaviour in terms of provocation, whether from the per-
spective of the bully or an outsider, is immaterial as a criterion of bullying (Parada 2006).
It has been argued that inclusion of this criterion carries the danger of creating a ‘blame
the victim’ culture, in which the actions of bullies are regarded as justifiable if provoked
(Parada 2006). This criterion appears to be based on observations that some victims were
also aggressive, or involved in bullying others, and so arguably may have provoked their
victimisation (Lee 2006). However, in the light of evidence of the reciprocal nature of
victimisation and bullying (see Marsh et al. 2004), the validity of this criterion must be
questioned.
In summary, the criteria of distress, repetition, intention to harm, power inequity and
provocation are common to the majority of definitions of bullying, but debate surrounds
the relevance, practical application and definition of each criterion. An important step in
informing the definitional debate, and the main aim of the present study, is an examination
of these definitional criteria and their appropriateness. This will be conducted by testing
the ability of the key criteria to predict a victimisation and a bullying summary score, each
of which reflects a combination of the frequency and variety of behaviours experienced
(or engaged in) in a six-month period. An early post-school population was selected for
the current investigation as it represents an under-researched population that can inform a
gap in our understanding of the developmental trajectory of bullying, as well as providing
a novel insight into the definitional framework of the bullying phenomenon within a non-
school population.

Method
Participants
Participants were 246 students enrolled in an introductory psychology course in their sec-
ond semester at the University of Sydney, Australia. There were 55 males and 188 females,
with gender not indicated for three. Ages ranged from 17 to 24 years (M = 18.61,
SD = 1.55). Participation was in partial fulfilment of course requirements. Ethical approval
was obtained from the University of Sydney Human Research Ethics Committee.
Emotional and Behavioural Difficulties 213

Measures
The Victimisation and Bullying Inventory
Exposure to and engagement in specific bullying acts was measured using the Victimisation
and Bullying Inventory (VBI). This was developed by Goldsmid (2011) for use with young
adults in post-school settings. It has been used in a large scale research project designed to
address the nature of involvement in victimisation and bullying behaviour across several
contexts, of which the present study forms a part. The instrument has been shown to have
good psychometric properties (Goldsmid and Howie 2013a), including predictive validity
(Goldsmid and Howie 2013b). It consists of 29 items that describe verbal, physical and
relational behaviours that could be perceived as bullying (e.g. I have been gossiped about;
I have verbally threatened someone). There is no explicit use of the term bullying. Each
item is presented once from the victim’s perspective and once, in a separate scale, from the
bully’s perspective, resulting in two 29-item scales.
A unique feature of the VBI is that it allows the simultaneous examination of
behaviours in multiple contexts. To do so, participants were asked to respond to all
29 victimisation and 29 bullying items first in relation to experiences in the tertiary con-
text, then the home, and finally the work context. These three contexts were selected on
the basis that they were judged to be highly frequented by the target early post-school
population.

Item format
Table 1 shows example items from the VBI scale. Each victimisation item first required
a yes/no endorsement regarding whether the respondent had experienced the behaviour
in the preceding six month period. Then for each endorsed item, the participant rated
frequency, victimisation distress, power inequity and provocation. The frequency rating
indicated how often the respondent had experienced the behaviour, on a five-point scale
(from 0 = Never to 4 = Almost Always). For distress, participants completed the state-
ment ‘When this happened it bothered me’ on a four-point scale (1 = Not at all to 4 = Very
Much). For power inequity, they completed the statement ‘I feel that the person who did
this is . . . powerful than me’ with the response alternatives of less (–1), about the same
(0), and more (1). For the provocation rating, they responded yes or no to the statement ‘I
feel that I did something to provoke this behaviour.’
Correspondingly, each endorsed bullying item required ratings of frequency, power
inequity and provocation. The distress rating was replaced by a rating of intention to harm,
which required participants to respond yes or no to the statement ‘When I did this I intended
to hurt or frighten the other person.’
For each of the three contexts, several scores were produced: an overall summary vic-
timisation score, an overall summary bullying score and a series of definitional criterion
scores, see Table 2.

Overall victimisation score


Each participant’s frequency ratings for the six-month period in the specified context were
summed across all 29 victimisation items, with a maximum possible score of 116. This
score therefore reflects a combination of the frequency and variety of behaviour experi-
enced, following the approach commonly used in the literature (Bendixen, Endresen, and
Olweus 2003; Parada 2000).
214

Table 1. Example victimisation and bullying items from the VBI.


Example Victimisation Item

I HAVE BEEN GOSSIPED OR BITCHED ABOUT Yes (1) No (0)


S. Goldsmid and P. Howie

THIS OCCURS: Never (0) Almost Never (1) Once in awhile (2) Often (3) Almost Always (4)
WHEN THIS OCCURRED IT BOTHERED ME: Not at all (1) A little bit (2) Quite a lot (3) Very much (4)
I FEEL THAT THE PERSON WHO DID THIS IS less (-1) about the same (0) more (1) POWERFUL THAN ME
I FEEL THAT I DID SOMETHING TO PROVOKE THIS BEHAVIOUR: Yes (1) No (0)

Example Bullying Item

I HAVE VERBALLY THREATENED SOMEONE Yes (1) No (0)


THIS OCCURS: Never (0) Almost Never (1) Once in awhile (2) Often (3) Almost Always (4)
WHEN I DID THIS I INTENDED TO HURT OR FRIGHTEN THE OTHER PERSON: Yes (1) No (0)
I FEEL THAT THE PERSON I DID THIS TO IS: less (1) about the same (0) more (-1) POWERFUL THAN ME
I FEEL THAT THE PERSON I DID THIS TO DID SOMETHING TO PROVOKE THIS BEHAVIOUR FROM ME: Yes (1) No (0)
Emotional and Behavioural Difficulties 215

Table 2. Summary description of scores generated by the VBI.


VBI Score Description

Victimisation score Sum of frequency ratings across all victimisation items


Bullying score Sum of frequency ratings across all bullying items
Victimisation distress score Average distress rating across all endorsed victimisation
items
Bullying intention to harm score Average intention rating across all endorsed bullying items
Victimisation power inequity Average power inequity rating across all endorsed
score victimisation items
Bullying power inequity score Average power inequity rating across all endorsed bullying
items
Victimisation provocation score Average provocation rating across all endorsed victimisation
items
Bullying provocation score Average provocation rating across all endorsed bullying
items

Overall bullying score


Corresponding to the victimisation score, each participant’s ratings of the frequency with
which they bullied others were summed across all 29 bullying items, with a maximum
possible score of 116.

Victimisation distress score


Each participant’s ‘bother’ ratings were summed across all victimisation items endorsed1
and then divided by the number of victimisation items endorsed, resulting in an average
distress score, with a range of 1 to 4.

Bullying intention to harm score


For each participant, the number of ‘yes’ responses (intention to harm) were summed
across all endorsed items, and then divided by the total number of endorsed items. This
resulted in a score with a range of 0 to 1, indicating the proportion of bullying behaviours
engaged in by the participant with an intention to harm their victim.

Victimisation power inequity


Participants’ ratings were summed across all endorsed victimisation items and then divided
by the number of endorsed items, resulting in an average score with a range of –1 to 1.
Scores below zero indicate an overall power imbalance in favour of the victim, and scores
above zero indicate an overall power imbalance in favour of the bully. Scores of zero indi-
cate that, on average, the endorsed behaviours were reported to occur between individuals
who were equal in power.

Bullying power inequity


Correspondingly, participants’ ratings of bullying power inequity were summed across all
endorsed items and then divided by the number of endorsed items. Scoring was reversed
216 S. Goldsmid and P. Howie

so that, consistent with Victimisation Power Inequity, scores below zero indicate an over-
all power imbalance in favour of the victim and scores above zero indicate an overall
power imbalance in favour of the bully, and scores of zero indicate no power imbalance
on average.

Victimisation provocation score


The number of ‘yes’ responses were summed, and then divided by the number of endorsed
victimisation items, reflecting the proportion of endorsed victimisation behaviours which
the victim felt they had provoked, with a range of 0 to 1.

Bullying provocation score


The number of ‘yes’ responses were summed, and then divided by the number of endorsed
bullying items. The resulting score represents the proportion of the total number of
endorsed bullying behaviours for which the bully felt that the victim had provoked the
behaviour, resulting in a score with a range of 0 to 1.

Procedure
After consenting to participate, students were sent online access instructions, and com-
pleted the VBI at a time convenient to them, using a home or campus computer, within a
two week period. Detailed instructions could be referred to by participants at any time. The
researcher was available for help via phone and email, but assistance was rarely requested.
To allow allocation of course credit, it was necessary for participants to enter their student
identification number and e-mail address, but in order to emphasise confidentiality, this
was done such that the information was clearly separate from the questionnaire. Students
were assured that all identifying information would be removed prior to data inspection
and stored separately.

Results
Correlations between overall victimisation score and definitional variables
Table 3 shows the means and standard deviations of all variables, and the correlations
between the victimisation and bullying scores and definitional variables, for each of the
tertiary, work and home contexts. Correlations between the overall victimisation score and
the victimisation definitional criterion variables were on the whole significant and mod-
erately strong. As predicted, overall victimisation scores were positively correlated with
victimisation distress, with the more frequent and varied the victimisation experiences, the
greater the distress reported.
As expected, power inequity showed modest positive associations with victimisation,
for the work and home contexts, with higher victimisation scores being associated with a
stronger belief that the bully was more powerful than themselves. There was no association
in the tertiary context. It should be noted, however, that participants, on average, reported
power inequity in favour of the bully for only a small proportion of victimisation behaviours
they reported experiencing (See Table 3).
Contrary to expectation, in all three contexts, victimisation scores were positively cor-
related with provocation scores, with higher victimisation associated with a stronger belief
Emotional and Behavioural Difficulties 217

Table 3. Means, standard deviations and correlation coefficients between the victimisation and
bullying scores and definitional variables for the tertiary, work, and home contexts.
Mean SD 2 3 4 5 6 7 8

Tertiary context
1 Victimisation score 7.52 7.72 .40∗∗ −.02 .26∗∗ .60∗∗ .12∗ .04 .17∗∗
2 Victimisation distress 1.91 1.08 − .24∗∗ .29∗∗ .25∗∗ .15∗ .16∗ .08
3 Victimisation power .02 .43 − −.06 −.07 −.02 −.10 −.08
inequity
4 Victimisation provocation .20 .28 − .20∗∗ .11 .02 .22∗∗
5 Bullying score 4.78 6.10 − .20∗∗ .13∗ .29∗∗
6 Bullying intention .68 .43 − .10 .38∗∗
7 Bullying power inequity −.07 .30 − .06
8 Bullying provocation .41 .42 −
Work context
1 Victimisation score 1.98 3.81 .63∗∗ .17∗∗ .33∗∗ .79∗∗ .44∗∗ .00 .44∗∗
2 Victimisation distress .85 1.13 − .31∗∗ .30∗∗ .43∗∗ .42∗∗
−.17∗∗ .42∗∗
3 Victimisation power .12 .41 − .09 .09 .17∗∗ −.36∗∗ .17∗∗
inequity
4 Victimisation provocation .07 .22 − .35∗∗ .37∗∗ −.15∗ .37∗∗
5 Bullying score 1.11 2.58 − .48∗∗ .06 .48∗∗
6 Bullying intention .20 .38 − −.10 1.00∗∗
7 Bullying power inequity .03 .32 − −.10
8 Bullying provocation .20 .38 −
Home context
1 Victimisation score 3.77 7.08 .60∗∗ .27∗∗ .37∗∗ .68∗∗ .35∗∗ −.23∗∗ .38∗∗
2 Victimisation distress .94 1.13 − .34∗∗ .53∗∗ .47∗∗ .46∗∗ −.19∗∗ .57∗∗
3 Victimisation power .08 .41 − .14∗ .19∗∗ .24∗∗ −.48∗∗ .19∗∗
inequity
4 Victimisation provocation .17 .30 − .42∗∗ .42∗∗ −.16∗ .38∗∗
5 Bullying score 4.56 8.34 − .33∗∗ −.13∗ .35∗∗
6 Bullying Intention .11 .25 − −.30∗∗ .50∗∗
7 Bullying Power Inequity .04 .33 − −.10
8 Bullying Provocation .29 .41 −

Notes: ∗ significant at the .05 level ;∗∗ significant at the .01 level.
Correlations between the various victimisation scores, and between the various bullying scores, are shown in bold,
and correlations between victimisation and bullying scores are shown in italics.

that the victim had provoked the behaviour. The average proportion of experienced victim-
isation behaviours reported as having provoked was .20, .07 and .17 in the tertiary, work
and home contexts respectively.

Correlations between overall bullying scores and definitional variables


Correlations between bullying scores and the bullying definitional criterion variables were
also generally significant (see Table 3). As predicted, bullying involvement was positively
and significantly correlated with intention to harm for all three contexts, with greater
involvement in bullying others associated with a more frequently reported intention to
harm the victim. On average, the proportion of bullying behaviours engaged in that partic-
ipants reported as having been intended to cause harm was .68, .20 and .11 in the tertiary,
work and home contexts respectively.
218 S. Goldsmid and P. Howie

The relationship between bullying and power inequity was less clear. Overall bullying
scores showed a small positive correlation with power inequity for the tertiary context
(r =.13, p <.05), suggesting that power inequity in favour of the bully was associated with
higher bullying scores, as expected. However, for the home context the association was in
the opposite direction, that is, power inequity in favour of the bully was associated with
lower bullying scores. There was no association for the work context. As for victimisation,
on average across all bullying items, participants reported power inequity in favour of the
bully for less than 9% of bullying behaviours engaged in.
Contrary to expectation, the overall bullying score was positively correlated with provo-
cation, with greater involvement in bullying others associated with a more frequent belief
that the victim provoked the behaviour. On average, the proportion of endorsed bullying
behaviours reported as provoked by the victim was .41, .20 and .29 for tertiary, work and
home respectively.

Overview of regression analyses


A series of hierarchical regression analyses were conducted to determine the contribution
of participants’ scores on the victimisation definitional criteria to the prediction of victim-
isation scores, after controlling for gender. Separate regression analyses were conducted to
predict tertiary-victimisation, work-victimisation and home-victimisation. For each analy-
sis, the predictor variables were entered in two blocks: in the first block, gender, and the
three victimisation definitional criterion variables (distress, power inequity and provoca-
tion), and in the second block, interactions between gender and each of the definitional
criterion variables.
For all regression analyses, gender was Cohen coded (male –.5, female .5), and pre-
dictor variables were mean-centred. One case with Cook’s distance greater than 1 was
excluded from the work bullying analysis (see Table 5) on the grounds that it could exert
undue influence on the model (Field 2009). The reported results will refer to the last model
to significantly increase the variance accounted for.

Regression: predicting victimisation scores in the tertiary, work and home contexts
from definitional variables
Table 4 shows the results of the regression analyses. R2 was significant for all three con-
texts, but only the work context showed a significant increase in the variance accounted
for when interactions with gender were added to the model. The following description
therefore applies to Model 1 for the tertiary and home contexts and Model 2 for the work
context. In the tertiary context, 23% of the variance in victimisation scores was accounted
for, and the corresponding percentages were 57% in the work context and 36% in the home
context.
Gender was a significant predictor for the tertiary and work contexts, with females
scoring, on average, 3.75 (tertiary) and 1.38 points (work) higher on overall victimisation
than males, when all definitional variables were controlled for. As expected, victimisation
distress was a significant predictor of the victimisation score in all three contexts, with
every point increase in victimisation distress corresponding to an increase in victimisa-
tion score between 2.07 and 3.32 points. For the work context only, there was a significant
interaction between distress and gender, with males showing a stronger association between
victimisation and distress (r =.80, p <.01) than females (r =.63, p <.01). That is, at any
Emotional and Behavioural Difficulties 219

Table 4. Predicting victimisation scores from definitional variables.


Tertiary Work Home

Variable b SE B β b SE B β b SE B β

Model 1
Gender −3.75 1.08 −.20∗∗ −1.07 .44 −.12∗∗ .10 .88 .01
Victimisation distress 2.81 .44 .39∗∗ 2.07 .18 .62∗∗ 3.32 .40 .53∗∗
Victimisation power −1.28 1.08 −.07 −.29 .47 −.03 1.40 .94 .08
inequity
Victimisation provocation 3.03 1.68 .11 2.49 .90 .14∗∗ 1.84 1.44 .08
Model 2
Gender −4.12 1.10 −.22∗∗ −1.38 .40 −.15∗∗ .02 .89 .00
Victimisation distress 3.39 .61 .47∗∗ 3.09 .21 .92∗∗ 3.52 .56 .56∗∗
Victimisation power −.17 1.55 −.01 −1.80 .56 −.20∗∗ 1.00 1.32 .06
inequity
Victimisation provocation 3.70 1.96 .13 1.86 .90 .10∗ 1.09 1.82 .05
Gender × victimisation −1.67 1.22 −.12 −3.22 .42 −.48∗∗ −.54 1.12 −.04
distress
Gender × Victimisation −2.99 3.10 −.08 5.05 1.11 .27∗∗ .86 2.65 .02
power inequity
Gender × victimisation −1.33 3.92 −.02 2.79 1.80 .08 2.47 3.64 .05
provocation

Notes: Tertiary: R2 = .23 (p < .01) for Step 1; R2 = .01 (p = ns) for Step 2. Work: R2 = .44 (p < .01) for Step
1; R2 = .13 (p < .01) for Step 2. Home: R2 = .36 (p < .01) for Step 1; R2 = .00 (p > .05) for Step 2
∗ significant at the .05 level; ∗∗ significant at the .01 level.

given level of victimisation females showed greater distress, with this discrepancy widen-
ing as victimisation level increased. Power inequity and provocation were significant
predictors of victimisation in the work but not the tertiary or home contexts. In the work
context, every point increase in provocation corresponded to a 1.86 point increase in work-
victimisation score (an association in the opposite direction to that expected). For power
inequity, the association with victimisation was qualified by an interaction with gender,
with females showing a significant association with victimisation in the expected direction
(greater inequity in favour of the bully associated with higher victimisation scores) and
males showing no association.

Regression: predicting bullying scores from definitional variables for behaviour in the
tertiary, work and home contexts
Corresponding regression analyses were run predicting tertiary-bullying, work-bullying
and home-bullying. Table 5 displays the results of the regression analyses. For the work
context, provocation was not included in the regression analysis because of its perfect
correlation with intention to harm (r = 1.0).
R2 was significant for all three contexts, but only in the work context was there a sig-
nificant increase in the amount of variance accounted for when interactions with gender
were added. Results for the relevant model will be reported for each context. The variance
in bullying scores accounted for was 12% in the tertiary context, 37% in the work context
and 16% in the home context.
Gender was a significant predictor of bullying in the tertiary and work contexts, with
females’ mean bullying scores 1.89 and .92 points lower than males in the tertiary and work
220 S. Goldsmid and P. Howie

Table 5. Predicting bullying scores from definitional variables.


Tertiary Work Home

Variable b SE B β b SE B β b SE B β

Model 1
Gender −1.89 .89 −.13∗ −.80 .30 −.15∗∗ −1.10 1.19 −.06
Intention to harm 1.26 .94 .09 3.18 .34 .53∗∗ 7.01 2.48 .20∗∗
Bullying power inequity −2.39 1.27 −.12 −.55 .40 −.08 .89 1.58 .04
Bullying provocation 3.67 .95 .25∗∗ 5.04 1.40 .25∗∗
Model 2
Gender −1.71 .90 −.12 −.92 .29 −.17∗∗ −.97 1.19 −.05
Intention to harm 2.49 1.23 .18∗ 4.75 .46 .79∗∗ 5.41 2.81 .16
Bullying power inequity 1.80 1.66 .09 2.70 .60 .37∗∗ −2.46 1.74 −.10
Bullying provocation 3.37 1.20 .23∗∗ 4.92 1.76 .24∗∗
Gender × Intention to −3.87 2.45 −.14 −4.16 .91 −.34∗∗ 6.60 5.61 .10
harm
Gender × Bullying 1.31 3.33 .03 −5.38 1.20 −.37∗∗ 7.18 3.49 .14∗
power inequity
Gender × bullying 1.12 2.39 .04 .34 3.53 .01
provocation

Notes: Tertiary: R2 = .12 (p < .01) for Step 1; R2 = .01 (p = ns) for Step 2. Work: R2 = .28 (p < .01) for
Step 1; R2 = .09 (p < .01) for Step 2. Home: R2 = .16 (p < .01) for Step 1; R2 = .02 (p > .05) for Step 2
∗ significant at the .05 level; ∗∗ significant at the .01 level
Provocation was not included in the work-bullying regression analysis because of its redundancy with intention
to harm (r = 1.0).
One participant was excluded from the work regression analysis with a Cook’s distance greater than 1

contexts respectively. Intention to harm was a significant predictor in the work and home
contexts, but not the tertiary context. The association was in the expected direction, with
every point increase in intention to harm corresponding to an increase of 4.75 (work) and
7.01 (home) points in engagement in bullying others. The significant interaction between
gender and intention to harm in the work context reflected the fact that the association
between intention and bullying score was stronger for females (r = .60, p < .01) than
for males (r = .50, p < .01). Power inequity was a significant predictor of bullying in
the work context only. Consistent with expectation, every point increase in power inequity
in favour of the bully corresponded to a 2.70 point increase in engagement in bullying.
There was a significant interaction with gender, with males (r = .29, p < .05) showing
more evidence of an association than females (r = −.07, p > .05). Provocation was a
significant predictor of bullying in the tertiary and home contexts, but as for workplace
victimisation, the direction of association was opposite to expectation, with every one point
increase in reported provocation predicting a 3.67 (tertiary) and 5.04 (home) point increase
in bullying.

Discussion
The aim of this study was to examine the relationships between overall victimisation and
bullying scores (calculated as the sum of frequency ratings across all endorsed behaviours)
and definitional criteria identified in the literature, using an early post-school sample.
Empirical exploration of definitional criteria is important as it directly impacts on the
future direction and integrity of bullying research. It was expected that, in line with
Emotional and Behavioural Difficulties 221

theory, victimisation scores would be predicted by victimisation distress, and victim per-
ception that the person bullying was more powerful, and that the bullying was unprovoked.
Similarly, it was expected that bullying scores would be predicted by intention to harm
on the part of the aggressor, and a bully perception that the object of bullying was less
powerful, and that the behaviour was unprovoked.

Victimisation distress and intention to harm


The regression analyses showed that, as expected based on previous research (e.g. Hawker
and Boulton 2000), degree of distress consistently predicted high victimisation scores, in
all three post-school contexts examined. This clearly supports the relevance of distress as
a definitional criterion of victimisation, and provides evidence that victimisation scores
generated by the VBI do, in accordance with definitions of bullying, reflect situations in
which the victim is distressed by the behaviour endured.
Further, in line with expectation, reports of intending to harm or frighten the victim
predicted higher bullying scores, at least in the work and home contexts. In the work and
home contexts, 20% and 11% respectively, of endorsed bullying behaviours were reported
as intended to cause harm. Thus, the positive association between intent and bullying indi-
cates that, on the rare occasions when intent was present, overall bullying was likely to
be severe. This supports intent as a definitional criterion in work and home contexts. The
lack of association in the tertiary context may reflect the fact that on average, the majority
(68%) of bullying behaviours endorsed in this context were reported as intended to cause
harm. This suggests that, when aggression occurs in this tertiary setting, intention to harm
is often present regardless of bullying severity. It is unlikely that the low home and work
percentages are entirely due to demand characteristics preventing honest reporting or to
limited self-awareness, since the same participants were prepared to acknowledge inten-
tion to harm in the tertiary context. However, the observed percentages are probably best
interpreted as reflecting a lower limit of the true situation.
A number of gender differences were observed for the work context. The positive asso-
ciation between victimisation and distress was stronger for males than females. That is,
overall, females reported higher levels of distress than males at any given level of victimi-
sation, and the gender discrepancy increased as victimisation and distress increased. This
finding is consistent with school-based findings that females are more distressed by victimi-
sation than males (Galen and Underwood 1997). Workplace research has generally failed to
find gender differences in overall level of distress or, more broadly, in mental health issues,
as a consequence of victimisation (Niedl 1996; Vartia and Hyyti 2002). Although in line
with the current finding, Vartia and Hyyti (2002) did report a trend towards a stronger
association between being bullied and mental ill-health issues for males than females.
A possible explanation for the gender differences in the present study may be that early
post-school individuals are more likely to hold junior workplace roles, compared with other
workplace research, which has typically sampled across the organizational hierarchy. Our
findings suggest that being in a position of vulnerability in the workplace may influence
or inflate gender differences in distress at high levels of victimisation. It is interesting that
in the current study, the gender difference in victimisation distress was not present in the
tertiary or home contexts (as discussed below).
Gender differences in the workplace were also found for the association between bul-
lying and intention to harm, but in this case, the association was stronger for females than
males. One interpretation of this is that males may engage in bullying-type behaviours at
work for a range of reasons in addition to intention to harm, for example, to seek peer
222 S. Goldsmid and P. Howie

acceptance. If so, the current findings suggest that bullying-type behaviours are used dif-
ferently by males and females in this population, but only in the work context, as similar
gender differences were not observed in the tertiary and home contexts. An alternate expla-
nation is that females are more self-aware than males, and thus more able to judge and
report the intent behind their actions. However, this is unlikely as it would have resulted
in similar gender differences within all three contexts. The intended purpose of engaging
in bullying-type acts, and in particular gender and contextual influences on readiness to
engage in such acts, is an interesting issue which has received very little research attention,
and would benefit from further empirical investigation.
It is important to note that intent to harm was positively correlated with provocation
in all three contexts, with a perfect correlation in the workplace. In light of this, the
obtained interaction between gender and intent may also be interpreted as females show-
ing a stronger association between provocation and bullying. That is, in the workplace,
females who bully others are more likely to feel that their behaviour was provoked by the
victim, and correspondingly are more likely to intend harm, than males engaging in sim-
ilar behaviours. This finding may again relate to the junior workplace positions held by
this early post-school sample. Fearing retribution or dismissal, females may have bullied
others only when they felt there was no alternative, that is, when highly provoked. This will
clearly require further exploration but highlights a gap in our understanding of the bullying
phenomenon in junior employees.

Power inequity
In the present study, females, but not males, reported higher levels of workplace victimi-
sation as power ratings moved towards perception of the bully as more powerful than the
victim. It is possible that early post-school females may be in an even less powerful work-
place position than males, increasing their vulnerability to victimisation. Lending weight
to this interpretation is the fact that similar gender differences for power inequity were not
observed within the tertiary or home contexts. Coupled with the current finding of work-
place gender differences in victimisation distress, this suggests that females may be more
vulnerable, not only to outcomes of victimisation, but also to victimisation itself, due to
their inferior power position within the workplace.
Gender differences were also found in the association between bullying and power
inequity, in the work, but not in the home or tertiary contexts. Unlike the association found
with victimisation, a modest association was found in males but no association in females.
In males, as expected, the level of bullying engaged in increased as the average power rat-
ings moved towards perception of the bully as more powerful than the victim. The finding
suggests that males are somewhat more likely to engage in bullying others when they per-
ceive they are more powerful than the victim(s). This gender difference was observed only
in the work context, again suggesting gender distinctions in power within the workplace
for this early post-school population.
Despite these workplace gender differences, overall the present results do not provide
strong support for a link between power inequity and bullying. Both participants who were
victimised and participants who bullied others reported that the behaviour most commonly
occurred between two equals, rather than between individuals with unequal power. This
finding is contrary to the majority of definitions of bullying. However, empirical support
for this definitional criterion is mixed, with some studies reporting support (e.g. Olweus
1993a) and others not (e.g. Ireland and Ireland 2003). Interestingly, the present findings
did show that, in all three contexts, the proportion of behaviours for which victimised
Emotional and Behavioural Difficulties 223

individuals reported that the bully was more powerful was almost double the proportion
for which bullies reported that they were more powerful. Given the importance of this
definitional criterion in delineating bullying behaviour from violence or general aggression
(Rigby 2001) this issue will need to be pursued. Further investigations may consider using
observational rather than self-report data, as it is not clear whether this finding reflects the
true situation or a lack of insight of the respondents.
One possible explanation for the low frequency of reports of power inequity in the
present study, may be that the measure (‘I feel that the person who did this to me/who
I did this to is less, about the same, more powerful than me’) did not adequately tap the
underlying construct referred to by bullying definitions, or was not sensitive enough to
capture small but important power inequities within the tertiary and home contexts. It may
also be that participants are less able to make accurate and objective power judgements in
reference to more abstract concepts such as social standing or physical attributes, or that
participant interpretations of power differed. As participants were not required to explain
the grounds on which they made power judgements in the current study, this issue cannot
be examined further here, but clearly merits further investigation.

Provocation
The present findings challenge the inclusion of provocation as a definitional element, show-
ing that individuals who bullied others blamed the victim for provoking the behaviour twice
as often as victimised individuals felt that they had provoked it. This ‘blame the victim’ ten-
dency was consistent across all three contexts examined. It is impossible to determine the
true frequency of provocation, but the small proportion of endorsed victimisation items
reported as provoked by the recipients themselves (.20, .07, and .17 in the tertiary, work
and home contexts respectively), may be more indicative of the individual’s vulnerability
and low self esteem (i.e. ‘I must be to blame’) than an accurate reflection of the cause of the
behaviour. On the other hand, the relatively high frequency with which those who bullied
reported provocation may in part reflect self justification, rather than the true cause of their
behaviour. Consistent with this, the correlations between provocation and level of bullying
were stronger in contexts where provocation was less frequently reported. That is, the few
who did report they were provoked by their victim(s) tended to be the ones that bullied
more extensively.
This study is the first to examine reports from both victimised and bullying individuals
regarding whether they felt specific behaviours were provoked, and thus the findings require
replication. Previous examinations of this issue have been confined to whether victimisa-
tion may be provoked by concurrent involvement in bullying by the victimised individual
(e.g. Unnever 2005). The current findings highlight a need to move beyond this simple
conceptualisation, and complement the growing evidence for the reciprocal effects theory
(Goldsmid and Howie 2013c; Marsh et al. 2004), which suggests a mutually reinforcing
relationship between victimisation and bullying involvement.
This study has a number of limitations. The use of self-reports may have resulted in
a level of social desirability in reporting, and shared method variance may have inflated
the associations. Furthermore, an undergraduate university sample cannot be assumed to
be representative of the wider early post-school population, in terms of educational level
and socioeconomic status. Although one of the strengths of this study is its contribution to
the very limited body of research on bullying in post-school populations, the nature of the
current sample means that caution is needed in generalising these findings to the greater
post-school population. It will also be important to determine whether similar definitional
224 S. Goldsmid and P. Howie

patterns are found in other populations, such as the community, school and workplace.
Finally, larger sample sizes with a more even gender ratio will be needed before conclusions
can be drawn with confidence regarding the potentially important gender differences noted
in this study.
In summary, the current findings have provided preliminary evidence of the construct
validity of the VBI. The findings also lend empirical support to a number of proposi-
tions regarding the criteria for bullying and victimisation. As predicted, in all contexts
studied, victimisation scores were positively associated with distress, and bullying scores
were positively associated with an intention to harm. Of note, the findings suggested
that, in the workplace, early post-school females may be more vulnerable than their male
counterparts, perhaps because they hold inferior power positions, resulting in increased
victimisation and distress. The current findings also provide some empirical support for
excluding provocation from definitions of bullying.

Note
1. For the purpose of all scores calculated, an item was considered to be endorsed if it was rated
as having occurred at or above a frequency of ‘almost never’ (1). This level was chosen as the
intention was to examine definitional relationships across the continuum of victimisation and
bullying (i.e. from infrequent to frequent behaviour), not only at high frequency levels.

References
Anderson, C. A., and B. J. Bushman. 2002. “Human Aggression.” Annual Review of Psychology 53:
27–51.
Baldry, A. C., and D. P. Farrington. 2004. “Evaluation of an Intervention Program for the Reduction
of Bullying and Victimisation in Schools.” Aggressive Behaviour 30: 1–15.
Bendixen, M., I. Endresen, and D. Olweus. 2003. “Variety and Frequency scales of Antisocial
Involvement: Which One is Better?.” Legal and Criminological Psychology 8: 13–150.
Connell, A., and D. P. Farrington. 1996. “Bullying among incarcerated young offenders: Developing
an interview schedule and some preliminary results.” Journal of Adolescence 19: 75–93.
Connolly, I., and M. O‘Moore. 2003. “Personality and family relations of children who bully.”
Personality and Individual Differences 35: 559–567.
Cowie, H., P. Naylor, I. Rivers, P. K. Smith, and B. Pereira. 2002. “Measuring workplace bullying.”
Aggressive and Violent Behaviour 7: 33–51.
Einarsen, S., H. Hoel, and G. Notelaers. 2009. “Measuring Exposure to Bullying and Harassment at
Work: Validity, Factor Structure and Psychometric Properties of the Negative Acts Questionnaire
– Revised.” Work and Stress 23 (1): 24–44.
Field, A. 2009. Discovering Statistics Using SPSS. London: SAGE Publications Ltd.
Galen, B. R., and M. K. Underwood. 1997. “A Developmental Investigation of Social Aggression
Among Children.” Developmental Psychology 33 (4): 589–600.
Goldsmid, S. 2011. “Bullying and Victimisation in an Early Post-School Population: An examination
of Prevalence, Factor Structure and Contextual Influences on Stability of Behaviours, using the
Victimisation and Bullying Inventory.” Unpublished doctoral dissertation, University of Sydney,
Australia.
Goldsmid, S., and P. Howie. 2013a. “Measuring Post-School Victimization and Bullying
Concurrently in Multiple Contexts: Development of the Victimisation and Bullying Inventory”.
Manuscript submitted for publication
Goldsmid, S., and P. Howie. 2013b. “Mediating or Moderating Role of Attributional Style in
the Association between Victimization and Psychological Distress.” Emotional & Behavioural
Difficulties. DOI: 10.1080/13632752.2013.803682
Goldsmid, S., and P. Howie. 2013c. “Relations Between Victimization and Engagement in Bullying
from One Context to Another Context.” Manuscript submitted for publication
Gregor, S. 2004. “Fighting Back Workplace Bullying in Australia.” In Psych 26 (5): 26–28.
Emotional and Behavioural Difficulties 225

Griffin, R. S., and A. M. Gross. 2004. “Childhood Bullying: Current Empirical Findings and Future
Directions for Research.” Aggressive and Violent Behaviour 9: 379–400.
Hawker, D. S. J., and M. J. Boulton. 2000. “Twenty Years’ Research on Peer Victimisation and
Psychosocial Maladjustment: A Meta-analytic Review of Cross-sectional Studies.” Journal of
Child Psychology and Psychiatry 41 (4): 441–455.
Ireland, J. L., and C. A. Ireland. 2003. “How Do Offenders Define Bullying? A Study of Adult, Young
and Juvenile Male Offenders.” Legal and Criminological Psychology 8: 159–173.
Lee, C. 2006. “Exploring Teacher’s Definitions of Bullying.” Emotional and Behavioural Difficulties
11 (1): 61–75.
Marsh, H. W., R. H. Parada, R. G. Craven, and L. Finger. 2004. “In the Looking Glass: A Reciprocal
Effect Model Elucidating the Complex Nature of Bullying, Psychological Determinants, and
the Central Role of Self-Concept”. In Bullying Implications for the Classroom, edited by C. E.
Sanders and G. D. Phye, pp. 63–107. San Diego, CA: Elsevier Academic Press.
Mynard, H., and S. Joseph. 2000. “Development of the Multidimensional Peer-Victimisation Scale.”
Aggressive Behaviour 26: 169–178.
Niedl, K. 1996. “Mobbing and Well-being: Economic and Personnel Development Implications.”
European Journal of Work and Organizational Psychology 5 (2): 239–249.
Olweus, D. 1993a. Bullying at School. What We Know and What We Can Do. Oxford: Blackwell
Publishers.
Olweus, D. 1993b. “Victimisation by Peers: Antecedents and Long-Term Outcomes”. In Social
Withdrawal, Inhibition, and Shyness in Childhood, edited by K. H. Rubin and J. B. Asendorpf.
Mahwah, NJ: Lawrence Erlbaum Associates Inc.
Parada, R. H. 2000. Adolescent Peer Relations Instrument: A Theoretical and Empirical Basis for the
Measurement of Participant Roles in Bullying and Victimisation of Adolescence: An Interim Test
Manual and a Research Monograph: A Test Manual. Sydney: University of Western Sydney.
Parada, R. H. 2006. “School Bullying: Psychosocial Determinants and Effective Intervention.”
Unpublished doctoral dissertation, University of Western Sydney, Sydney.
Rigby, K. 1997. “What Children Tell Us about Bullying in Schools.” Children Australia 22 (2):
26–34.
Rigby, K. 2001. “Bullying in Schools and in the Workplace”. In Bullying From Backyard to
Boardroom, edited by P. McCarthy, J. Rylance, R. Bennett, and H. Zimmerman. Marrickville:
The Federation Press.
Roland, E. 2002. “Aggression, Depression, and Bullying Others.” Aggressive Behaviour 28:
198–206.
Salmivalli, C., and E. Nieminen. 2002. “Proactive and Reactive Aggression Among School Bullies,
Victims, and Bully-Victims.” Aggressive Behaviour 28: 30–44.
Sanders, C. E. 2004. “What Is Bullying?”. In Bullying Implications for the Classroom, edited by C.
E. Sandersand G. D. Phye, pp. 1–16. San Diego, CA: Elsevier Academic Press.
Slee, P., and K. Rigby. 1993. “The relationship of Eysenck’s Personality Factors and Self-Esteem to
Bully-Victim Behaviour in Australian Schoolboys.” Personality and Individual Differences 14
(2): 371–373.
Smith, P. K., and P. Brain. 2000. “Bullying in Schools: Lessons From Two Decades of Research.”
Aggressive Behaviour 26: 1–9.
Unnever, J. D. 2005. “Bullies, Aggressive Victims, and Victims: Are They Distinct Groups?.”
Aggressive Behavior 31: 153–171.
Vartia, M., and J. Hyyti. 2002. “Gender Differences in Workplace Bullying Among Prison Officers.”
European Journal of Work and Organizational Psychology 11 (1): 113–126.
Copyright of Emotional & Behavioural Difficulties is the property of Routledge and its
content may not be copied or emailed to multiple sites or posted to a listserv without the
copyright holder's express written permission. However, users may print, download, or email
articles for individual use.

You might also like